[pymvpa] How can I obtain the sensitivity map for crossvalication of selected features using RFE?

[Meng Liang]

Dear all,

I'm trying to combine RFE and cross-validation but not sure how to obtain the final sensitivity map. I used the example codes on the following webpage: 

http://www.pymvpa.org/generated/mvpa2.featsel.rfe.RFE.html

and then put the above classifier (clf = FeatureSelectionClassifier) in CrossValication:

cv = CrossValidation(clf, NFoldPartitioner(), enable_ca=['confusion','stats'])

Is it correct? If so, how can I obtain the final sensitivity map of the selected features? I tried the following:

sen = clf.get_sensitivity_analyzer()
cv_sen = RepeatedMeasure(sen, NFoldPartitioner())
error = cv(ds)
sensmap_cv = cv_sen(ds)

But I always got the following error message:
"RuntimeError: Cannot reverse-map data since the original data shape is unknown. Either set `dshape` in the constructor, or call train()."

Any help would be much appreciated!

Best,
Meng
 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20130120/b0bd73ad/attachment.html>